1
0
mirror of https://github.com/l1ving/youtube-dl synced 2025-02-09 13:13:06 +08:00

franceinter download video when available (ignoring audio)

This commit is contained in:
Eitan Postavsky 2018-02-25 09:36:42 -05:00 committed by unknown
parent 19a352854f
commit cc504ab6c1

View File

@ -1,6 +1,8 @@
# coding: utf-8
from __future__ import unicode_literals
import re
from .common import InfoExtractor
from ..utils import month_by_name
@ -8,26 +10,53 @@ from ..utils import month_by_name
class FranceInterIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?franceinter\.fr/emissions/(?P<id>[^?#]+)'
_TEST = {
'url': 'https://www.franceinter.fr/emissions/affaires-sensibles/affaires-sensibles-07-septembre-2016',
'md5': '9e54d7bdb6fdc02a841007f8a975c094',
'info_dict': {
'id': 'affaires-sensibles/affaires-sensibles-07-septembre-2016',
'ext': 'mp3',
'title': 'Affaire Cahuzac : le contentieux du compte en Suisse',
'description': 'md5:401969c5d318c061f86bda1fa359292b',
'upload_date': '20160907',
_TESTS = [
{
'url': 'https://www.franceinter.fr/emissions/affaires-sensibles/affaires-sensibles-07-septembre-2016',
'md5': '9e54d7bdb6fdc02a841007f8a975c094',
'info_dict': {
'id': 'affaires-sensibles/affaires-sensibles-07-septembre-2016',
'ext': 'mp3',
'title': 'Affaire Cahuzac : le contentieux du compte en Suisse',
'description': 'md5:401969c5d318c061f86bda1fa359292b',
'upload_date': '20160907',
},
},
}
{
'note': 'Audio + video (Dailymotion embed)',
'url': 'https://www.franceinter.fr/emissions/l-instant-m/l-instant-m-13-fevrier-2018',
'info_dict': {
'id': 'x6eow0v',
'ext': 'mp4',
'title': 'Propagande, stéréotypes, spectaculaire : les jeux vidéo font-ils du mal à l\'Histoire ?',
'description': 'Regardez Propagande, stéréotypes, spectaculaire : les jeux vidéo font-ils du mal à l\'Histoire ? par France Inter sur Dailymotion',
'uploader_id': 'x2q2ez',
'timestamp': 1518516881,
'uploader': 'France Inter',
'upload_date': '20180213',
},
},
]
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
video_url = self._search_regex(
# If there is a video version, use that instead
maybe_video_uuid = re.search(r'data-video-anchor-target=["\']([^"\']+)', webpage)
if maybe_video_uuid:
video_uuid = maybe_video_uuid.group(1)
video_url = self._search_regex(
r'(?sx)data-uuid=["\']%s.*?<iframe[^>]*src=["\']([^"\']+)' % video_uuid,
webpage, 'video url', fatal=False, group=1)
if video_url:
return self.url_result(video_url)
audio_url = self._search_regex(
r'(?s)<div[^>]+class=["\']page-diffusion["\'][^>]*>.*?<button[^>]+data-url=(["\'])(?P<url>(?:(?!\1).)+)\1',
webpage, 'video url', group='url')
webpage, 'audio url', group='url')
title = self._og_search_title(webpage)
description = self._og_search_description(webpage)
@ -50,7 +79,7 @@ class FranceInterIE(InfoExtractor):
'description': description,
'upload_date': upload_date,
'formats': [{
'url': video_url,
'url': audio_url,
'vcodec': 'none',
}],
}